Below is an email exchange between Nanex and William Alpert, Barron's Senior Editor and author of the article that was the subject of our paper: Robber Barrons. Alpert initiated the email, and we responded.
Michael Lewis, after reading this exchange, was surprised at Alpert's assertion that Barron's could not reach him, commenting: "Funny. I've never had a publication "fail to reach me" without even knowing they had tried."
Our main issue with Alpert's article is that it never states that Marketable Limit orders were excluded (a big deal as explained here). Alpert informs us that this information does exist in the code which was supplemental to the story. But the reader would have to download and install software, then download, compile and run his code to discover that fact. We think that is entirely unreasonable.
On Wed, Mar 4, 2015 at 3:43 PM, Alpert, William wrote:
[You read some of my code] ..and for that, I’m thankful. If you’d read the story more carefully, however, you would have noticed us clearly saying we were analyzing market orders.You present that fact as if you’d uncovered a gotcha after diligent investigation.Clearly? No. Not clear at all.
I ran this by quite a few people. You were not clear about excluding Marketable Limits in your article, in any way. Marketable Limit orders, to many people, especially those who view the stock market as a "Modern Market", equate Marketable Limit orders to Market orders with an added safety feature (the limit) in case of a flash crash (this was a topic of discussion in the months after the flash crash: using Marketable Limits as a substitute for Market orders).
So no, your article was not clear about that, even under a microscope. Even in hindsight: right now, as it stands it is not clear. Consider a retraction, or at least an apology to your readers. They were definitely deceived.
Our story was about comparing firms’ executions. The market orders allow the most apples-to-apples comparison, with a sample size that’s comfortably larger than the single stock selections that you present to journalists. The market makers handle and report limit orders differently, depending on how they handle IOCs – confounding a comparison.Quoting your opening sentence:
In the furor surrounding last year’s best-seller by Michael Lewis, many retail investors were spooked by the book’s claim that high-frequency traders use their technology edge to pick off the little guys, who, the author claims, were “easy kill” for the professionals.And what is this about?
Our attempts to reach Lewis through his publisher went unanswered.
When asked about Barron's attempt to contact him, Michael Lewis told us:
"Funny. I've never had a publication "fail to reach me" without me even knowing they had tried."
What exactly would you have asked Michael if you reached him?
"Hey M. Lewis, if we exclude 90% of the data, and pay no attention to the Wholesalers front running those posting limit orders on lit markets, then _____ ?"
Seriously, what is your question?
I hope you ran my code all the way, because if you did you’d see a browser-readable report that tells you, readers and the world a bunch of things we lacked space to detail in the magazine story – including a nice chart that prominently contrasts the price improvement on market orders with that on marketable limit orders, for whichever 605-reporting firm you choose to study….averaged across all stocks, however, and not just a single name like Apple.Yes, I saw those. Long after reading your article. After I found the comments on Marketable Limits, and before I began writing that piece, I asked quite a few people, many who are industry experts, if they thought Barron's discarded Marketable Limit orders in the article.
Q: Guess how many said yes?
A: ZERO.
Q: Guess how many were surprised?
A: ALL OF THEM.
What you are implying is that people who read your article need to download a 3rd party code package, compile and run it to realize that the *majority* of orders were excluded - orders that are central and relevant to retail traders. Orders that, had you actually studied, should have shown you, clearly, that Michael Lewis was SPOT ON in how retail investors ARE "easy kill" for professionals.
I truly hope you were bamboozled, because the alternative is fraud. The case for a simple misunderstanding is weak.
Marketable limit orders, indeed limit orders generally, seem like a great topic for subsequent stories, don’t you think ?Yes.
An even better topic: why are more than 25% of all price improved shares only price improved by the minimum amount $0.0001 ?
That was the #1 question you should have asked all the Wholesalers for your piece. More on that: http://www.nanex.net/aqck2/
3519.html Since we’re on the subject of straight-talk, you might explain why it makes any sense to talk about including/excluding the inside/at/near the quote limit orders in study that’s measuring price improvement. Your chart might mislead a reader who’s unwitting.Are you really asking this? Those other order types, at least for Citadel - the one you crowned king, NEVER HAVE PRICE IMPROVEMENT! This is probably another one of those questions you should have asked the Wholesalers.
Furthermore, those other order types represent a tiny fraction of total orders. There's a chart on the page I posted. No need to download, compile and run any code.
Nanex Research
Inquiries: pr@nanex.net